Breastfeeding is Associated with Reduced Childhood Hospitalization: Evidence from a Scottish Birth Cohort (1997-2009) Omotomilola M. Ajetunmobi, MSc1, Bruce Whyte, MSc2, James Chalmers, MRCGP, FFPH1,3, David M. Tappin, MRCP(UK)4, Linda Wolfson, MPH5, Michael Fleming, MSc1, Alison MacDonald, MPH6, Rachael Wood, PhD1, and Diane L. Stockton, PhD1, on behalf of the Glasgow Centre for Population Health Breastfeeding Project Steering Group* Objective To evaluate the risk of childhood hospitalization associated with infant feeding patterns at 6-8 weeks of age in Scotland. Study design A retrospective population level study based on the linkage of birth, death, maternity, infant health, child health surveillance, and admission records for children born as single births in Scotland between 1997 and 2009 (n = 502 948) followed up to March 2012. Descriptive analyses, Kaplan Meier tests, and Cox regression were used to quantify the association between the mode of infant feeding and risk of childhood hospitalization for respiratory, gastrointestinal, and urinary tract infections, and other common childhood ailments during the study period. Results Within the first 6 months of life, there was a greater hazard ratio (HR) of hospitalization for common childhood illnesses among formula-fed infants (HR 1.40; 95% CI 1.35-1.45) and mixed-fed infants (HR 1.18; 95% CI 1.11-1.25) compared with infants exclusively breastfed after adjustment for parental, maternal, and infant health characteristics. Within the first year of life and beyond, a greater relative risk of hospitalization was observed among formula-fed infants for a range of individual illnesses reported in childhood including gastrointestinal, respiratory, and urinary tract infections, otitis media, fever, asthma, diabetes, and dental caries. Conclusions Using linked administrative data, we found greater risks of hospitalization in early childhood for a range of common childhood illnesses among Scottish infants who were not exclusively breastfed at 6-8 weeks of age. (J Pediatr 2015;166:620-5).
B
reastfeeding enhances child health and development, with the potential to give every child a healthy start.1,2 In developed countries however, there remains continued debate on the size of health benefit based on the available evidence, which has been limited by methodologic issues related to sample size, quality of data, or adjustment for confounding factors3-6; particularly socioeconomic factors associated with both the choice and duration of infant feeding and child health outcomes.3,4 As in the rest of the United Kingdom, increasing rates of childhood hospitalizations have been observed in Scotland, particularly for acute infections among infants.7,8 These hospitalizations may contribute to substantial savings in the health service if breastfeeding rates increased marginally.9 Current trends in Scotland, however, show relatively stable exclusive breastfeeding rates and an increasing proportion of mixed (formula and human milk) fed infants. Approximately one-half of infants born annually initiate exclusive breastfeeding, decreasing to 25% by the review at 6-8 weeks after birth.10 Based on linkage of administrative data for a population cohort of Scottish born infants, we describe patterns of hospitalization observed in early childhood in relation to the mode of infant feeding reported at 6-8 weeks after birth, adjusted for a range of socioeconomic factors. The analyses focus on hospitalization for a range of conditions in which breastfeeding has been shown to be protective6,11-13 and, additionally, on conditions frequently reported in the cohort.
Methods This was a retrospective cohort study of singleton births in Scotland between 1997 and 2009 using anonymized extracts of linked administrative data provided by the Information Services Division, National Health Service National Services Scotland. Approval for the project design and confidentiality of patient data was provided by the Privacy Advisory Committee of National Health Service
From the 1Information Services Division, National Health Service, National Services Scotland; 2Glasgow Centre for Population Health, Glasgow, Scotland; 3National Education Scotland, National Health Service, National Services Scotland, Edinburgh, Scotland; 4Pediatric Epidemiology and Community Health (PEACH) Unit, Glasgow University; 5National Health Service (NHS), Greater Glasgow and Clyde, Glasgow, Scotland; and 6 National Health Service (NHS) Health Scotland
*Additional members of the Glasgow Centre for Population Health Breastfeeding Project Steering Group are available at www.jpeds.com (Appendix). Supported by the Scottish Collaboration of Public Health Research and Policy (SCPH/08 and SCPH/17) and the Glasgow Centre for Population Health. The authors declare no conflicts of interest. 0022-3476/Copyright ª 2015 The Authors. Published by Elsevier Inc.
HR PAF
Hazard ratio Population-attributable fraction
This is an open access article under the CC BY-NC-ND license (http:// creativecommons.org/licenses/by-nc-nd/3.0/) http://dx.doi.org/10.1016/j.jpeds.2014.11.013
620
Vol. 166, No. 3 March 2015 National Services Scotland, a body set up to ensure the appropriate use of patient identifiable information.14 Further ethical permission was not required. The records were linked in 2 phases via a combination of probabilistic matching techniques and the use of the Community Health Index, a unique identifier developed for health records in Scotland.15 Phase 1 comprised linkage of births, deaths, migration, maternity, infant health, and child health surveillance review records,16 which was extended in Phase 2 to include episodes of hospital admission from birth until March 2012. Thus, each child in the cohort could be followed up from birth for at least 2.25 years and up to 15 years (depending on their birth year). Infants with a diagnosis of congenital anomaly, a condition originating in the perinatal period, with invalid infant feeding records, and of non-Scottish residents were excluded from analysis. Outcome Variables The main outcome examined was a primary discharge diagnosis for selected conditions from a review of literature, including gastrointestinal infections, lower and upper respiratory tract infections, otitis media, asthma, urinary tract infections, allergy, eczema, and diabetes, and conditions frequently reported in the cohort such as fever and dental caries (Table I; available at www.jpeds.com). Definition of Infant Feeding Infant feeding reported at the 6-8 weeks of age review, defined as the predominant mode of infant feeding on the day preceding data collection, was divided into 3 categories: “exclusive breastfeeding,” “formula feeding,” and “mixedbreast and formula feeding.” Statistical Analyses Descriptive and univariate (Kaplan Meier curves) analyses were used to identify variables associated with hospital admission and infant feeding. Multivariate analyses (Cox regression analyses) were conducted to quantify the independent contribution of infant feeding at the 6-8 week review on hospitalization in childhood. The Cox models included only variables significantly associated with the health outcome from the univariate analyses. Each model was tested for proportionality over time, adjustments were made to ensure the best fit, and violations noted. Adjustment was made for parental factors, delivery and infant health characteristics, and features of the health care system.16 In addition, area deprivation derived using postcode at birth based on the Scottish Index of Multiple Deprivation (200617) and maternal ethnic and religious background derived from the mother’s given name (ie, OnoMAP18) were included in the analyses. Additional analysis was conducted to assess the risk of hospitalization for injuries (a condition not causally associated with breastfeeding), to test the adequacy of adjustment for socioeconomic and demographic confounders included in the linked dataset.
A series of models were applied over varying periods of follow-up (using STATA vs11; StataCorp LP, College Station, Texas) to estimate the risk of hospital admission associated with infant feeding patterns: up to 6 months (that reflects the recommended duration of exclusive breastfeeding), 6-27 months (point of “equal” follow-up for all infants in the cohort), and up to 15 years (full-follow-up). An estimate of the time from birth to event (ie, first hospital admission) was derived from the merged datasets and a marker applied to infants who had a hospital admission. Those who had migrated or died before the end of the observation period (March 2012) were censored from follow-up at the point of migration/death. The variables were entered into the model iteratively ie, infant feeding at the 6- to 8-week review was entered first, followed by parental/background variables, then delivery and infant health variables. The model entry significance was 0.05. Population-attributable fractions (PAFs) were used to quantify the number of new cases that may have been avoided among formula fed children in each model using the formula: PAF = [(hazard ratio [HR] 1)/HR] proportion of the exposed population.
Results Of the 502 948 singletons born between 1997 and 2009 included in the analysis, 63% were born by spontaneous/ normal delivery, 8% had teenage mothers, 17% were born to single parents or parents living apart, 45% were born to first-time mothers, 22% had mothers who smoked, and 27% were residents in the most socioeconomically deprived areas at birth (Table II; available at www.jpeds.com). By the review at 6-8 weeks of age, 27% of infants were reported as exclusively breastfeeding, 9% as having mixed feeding, and 64% as formula-feeding. During the study period, 137 905 (27%) of the infants had been hospitalized at least once for any of the selected conditions. At the first recorded hospital event, 31% were younger than 1 year, 29% were 1-2 years, 19% were 3-5 years, and 21% were aged 5 years or older. Most of the first hospital events were “emergency admissions” (75%), especially among infants
B
reastfeeding enhances child health and development, with the potential to give every child a healthy start.1,2 In developed countries however, there remains continued debate on the size of health benefit based on the available evidence, which has been limited by methodologic issues related to sample size, quality of data, or adjustment for confounding factors3-6; particularly socioeconomic factors associated with both the choice and duration of infant feeding and child health outcomes.3,4 As in the rest of the United Kingdom, increasing rates of childhood hospitalizations have been observed in Scotland, particularly for acute infections among infants.7,8 These hospitalizations may contribute to substantial savings in the health service if breastfeeding rates increased marginally.9 Current trends in Scotland, however, show relatively stable exclusive breastfeeding rates and an increasing proportion of mixed (formula and human milk) fed infants. Approximately one-half of infants born annually initiate exclusive breastfeeding, decreasing to 25% by the review at 6-8 weeks after birth.10 Based on linkage of administrative data for a population cohort of Scottish born infants, we describe patterns of hospitalization observed in early childhood in relation to the mode of infant feeding reported at 6-8 weeks after birth, adjusted for a range of socioeconomic factors. The analyses focus on hospitalization for a range of conditions in which breastfeeding has been shown to be protective6,11-13 and, additionally, on conditions frequently reported in the cohort.
Methods This was a retrospective cohort study of singleton births in Scotland between 1997 and 2009 using anonymized extracts of linked administrative data provided by the Information Services Division, National Health Service National Services Scotland. Approval for the project design and confidentiality of patient data was provided by the Privacy Advisory Committee of National Health Service
From the 1Information Services Division, National Health Service, National Services Scotland; 2Glasgow Centre for Population Health, Glasgow, Scotland; 3National Education Scotland, National Health Service, National Services Scotland, Edinburgh, Scotland; 4Pediatric Epidemiology and Community Health (PEACH) Unit, Glasgow University; 5National Health Service (NHS), Greater Glasgow and Clyde, Glasgow, Scotland; and 6 National Health Service (NHS) Health Scotland
*Additional members of the Glasgow Centre for Population Health Breastfeeding Project Steering Group are available at www.jpeds.com (Appendix). Supported by the Scottish Collaboration of Public Health Research and Policy (SCPH/08 and SCPH/17) and the Glasgow Centre for Population Health. The authors declare no conflicts of interest. 0022-3476/Copyright ª 2015 The Authors. Published by Elsevier Inc.
HR PAF
Hazard ratio Population-attributable fraction
This is an open access article under the CC BY-NC-ND license (http:// creativecommons.org/licenses/by-nc-nd/3.0/) http://dx.doi.org/10.1016/j.jpeds.2014.11.013
620
Vol. 166, No. 3 March 2015 National Services Scotland, a body set up to ensure the appropriate use of patient identifiable information.14 Further ethical permission was not required. The records were linked in 2 phases via a combination of probabilistic matching techniques and the use of the Community Health Index, a unique identifier developed for health records in Scotland.15 Phase 1 comprised linkage of births, deaths, migration, maternity, infant health, and child health surveillance review records,16 which was extended in Phase 2 to include episodes of hospital admission from birth until March 2012. Thus, each child in the cohort could be followed up from birth for at least 2.25 years and up to 15 years (depending on their birth year). Infants with a diagnosis of congenital anomaly, a condition originating in the perinatal period, with invalid infant feeding records, and of non-Scottish residents were excluded from analysis. Outcome Variables The main outcome examined was a primary discharge diagnosis for selected conditions from a review of literature, including gastrointestinal infections, lower and upper respiratory tract infections, otitis media, asthma, urinary tract infections, allergy, eczema, and diabetes, and conditions frequently reported in the cohort such as fever and dental caries (Table I; available at www.jpeds.com). Definition of Infant Feeding Infant feeding reported at the 6-8 weeks of age review, defined as the predominant mode of infant feeding on the day preceding data collection, was divided into 3 categories: “exclusive breastfeeding,” “formula feeding,” and “mixedbreast and formula feeding.” Statistical Analyses Descriptive and univariate (Kaplan Meier curves) analyses were used to identify variables associated with hospital admission and infant feeding. Multivariate analyses (Cox regression analyses) were conducted to quantify the independent contribution of infant feeding at the 6-8 week review on hospitalization in childhood. The Cox models included only variables significantly associated with the health outcome from the univariate analyses. Each model was tested for proportionality over time, adjustments were made to ensure the best fit, and violations noted. Adjustment was made for parental factors, delivery and infant health characteristics, and features of the health care system.16 In addition, area deprivation derived using postcode at birth based on the Scottish Index of Multiple Deprivation (200617) and maternal ethnic and religious background derived from the mother’s given name (ie, OnoMAP18) were included in the analyses. Additional analysis was conducted to assess the risk of hospitalization for injuries (a condition not causally associated with breastfeeding), to test the adequacy of adjustment for socioeconomic and demographic confounders included in the linked dataset.
A series of models were applied over varying periods of follow-up (using STATA vs11; StataCorp LP, College Station, Texas) to estimate the risk of hospital admission associated with infant feeding patterns: up to 6 months (that reflects the recommended duration of exclusive breastfeeding), 6-27 months (point of “equal” follow-up for all infants in the cohort), and up to 15 years (full-follow-up). An estimate of the time from birth to event (ie, first hospital admission) was derived from the merged datasets and a marker applied to infants who had a hospital admission. Those who had migrated or died before the end of the observation period (March 2012) were censored from follow-up at the point of migration/death. The variables were entered into the model iteratively ie, infant feeding at the 6- to 8-week review was entered first, followed by parental/background variables, then delivery and infant health variables. The model entry significance was 0.05. Population-attributable fractions (PAFs) were used to quantify the number of new cases that may have been avoided among formula fed children in each model using the formula: PAF = [(hazard ratio [HR] 1)/HR] proportion of the exposed population.
Results Of the 502 948 singletons born between 1997 and 2009 included in the analysis, 63% were born by spontaneous/ normal delivery, 8% had teenage mothers, 17% were born to single parents or parents living apart, 45% were born to first-time mothers, 22% had mothers who smoked, and 27% were residents in the most socioeconomically deprived areas at birth (Table II; available at www.jpeds.com). By the review at 6-8 weeks of age, 27% of infants were reported as exclusively breastfeeding, 9% as having mixed feeding, and 64% as formula-feeding. During the study period, 137 905 (27%) of the infants had been hospitalized at least once for any of the selected conditions. At the first recorded hospital event, 31% were younger than 1 year, 29% were 1-2 years, 19% were 3-5 years, and 21% were aged 5 years or older. Most of the first hospital events were “emergency admissions” (75%), especially among infants
